Available online at www.sciencedirect.

com

Procedia CIRP 7 (2013) 401 – 406

Forty Sixth CIRP Conference on Manufacturing Systems 2013

Statistical Process Control as a Service: An Industrial Case Study

a a,*
, , Peter Butalaa, Alojzij Slugaa
a
University of Ljubljana, Department of Manufacutirng Systems and Cont -1000 Ljubljana, Slovenia

* Corresponding author. Tel.: +386-1-4771-744. E-mail address: rok.vrabic@fs.uni-lj.si.

Abstract

Strong operations support is one of the key requirements for competitive success of modern manufacturing organisations. An
important aspect of operations support is Statistical Process Control (SPC); the use of statistical methods for monitoring and control
of manufacturing processes and products. However, implementation of SPC requires a certain amount of statistical knowledge and
understanding. Although this is not an issue for big companies (e.g. in automotive sector), smaller companies are unable to provide
the required knowledge in-house. In this paper, a service-driven approach for SPC is proposed, in which SPC is outsourced through
the use of modern information and communication technologies, such as web services. This Statistical Process Control as a Service
approach is illustrated and discussed through an industrial case study.

© 2013 The
© 2013 TheAuthors.
Authors.Published
Publishedbyby Elsevier
Elsevier B.V.
B.V. Open access under CC BY-NC-ND license.
Selection and/or
Selection and peer-review
peer-review underunder responsibility
responsibility of Professor
of Professor PedrodoFilipe
Pedro Filipe CarmodoCunha
Carmo Cunha

Keywords: Quality assurance; Statistical process control; Service oriented architecture

integrated solutions, smaller companies frequently resort

1. Introduction
Statistical Process Control (SPC) is a powerful
collection of problem-solving tools useful in achieving
manufacturing process stability and improving capability
through the reduction of variability [1]. It may be used
when a large number of similar items is being produced.
The underlying assumption is that good items are
produced when processes are in control with respect to
target values. The main objective of SPC is to give a
signal when the process changes, i.e. its mean moves
away from the target value and/or its variability
increases.
In essence, SPC is used to support manufacturing
operations. When a signal shows that the process is
changing, it is the up to the machine operator to trigger a
corrective action. Traditionally, the most important SPC
tool is control chart, which graphically represents
process data and shows whether the process is under
statistical control or not.
Implementation of SPC requires a certain amount of
statistical knowledge and understanding, and is often
integrated with commercial SCADA, MES, and/or ERP
software. While bigger companies are able to afford
to manual methods, or are even left without SPC [2-4].
Furthermore, development of custom solutions is
expensive and requires expert knowledge which smaller
companies are unable to provide in-house.
In this paper, a service-driven approach for SPC is
proposed. It is shown how modern information and
communication technology (ICT) solutions can be used
to build a distributed electronic SPC system (eSPC). The
core of the system is a web service, which provides the
means for remotely generating SPC reports such as
control charts. The architecture of the system is
presented along with an implementation of a user
interface. The system is demonstrated in an industrial
case study of plastic tube production.
The remainder of the paper is structured as follows.
In section 2, a literature review of SPC, control charts,
and service-oriented architecture (SOA) in
manufacturing is presented. Section 3 introduces the
proposed approach of SPC as a Service, including the
system architecture and the inner workings of the
system. Section 4 presents the case study and section 5
concludes the paper.

2212-8271 © 2013 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of Professor Pedro Filipe do Carmo Cunha
doi:10.1016/j.procir.2013.06.006
402 Gašper Škulj et al. / Procedia CIRP 7 (2013) 401 – 406

2. Literature review STAKEHOLDERS NETWORK

SET OF SET OF
OBJECTIVES & RULES &
In this section, two main topics are reviewed. The CONSTRAINTS CONVENTIONS
first one is SPC with control charts and the second one is

ENVIROMENT

ENVIROMENT
SOA in manufacturing. Together, they form basis for
SPC as a Service concept, presented in section 3.
The main reason for implementation of SPC in SERVICE SERVICE SERVICED
manufacturing processes is the need for higher and OBJECT PROCESS OBJECT
constant quality. Quality can be seen in various
dimensions; level of performance, reliability, durability, SUBJECT
RESOURCES
serviceability, aesthetic, features, perceived quality, and
conformance to standards [5]. This is the traditional way SERVICES
to describe quality and is summed up as eans SERVICE UNITS

. In contrast to the traditional definition

that is consumer oriented, the modern
1], is primarily
concerned with manufacturing aspects [6]. An important
interpretation of the mo
improvement is the reduction of variability in processes
[1]. In this view, SPC is seen as a
mechanism for controlling variability.
In SPC the process is observed as a system with a set
of inputs and outputs, and is controlled according to
quality-related specifications set during product
development. It is important to recognise that product
quality is strongly related to the manufacturing process.
To produce high quality outputs, process parameters are
set specific values that correspond to nominal or target
values. The actual values can differ from nominal values
in the predetermined area between an upper specification
limit (USL) and a lower specification limit (LSL).
Control charts are one of the primary and well known
techniques of SPC. A systematic use of control charts is
intended to reduce variability and therefore, according to
the definition, improve process quality. Many types of
control charts are known, differing with respect to
specifics of the controlled phenomena. Control charts
can be roughly divided into control charts for variables
and control charts for attributes [1].
Although the use of control charts is fairly simple, it
is not always done correctly and systematically. In the
light of SOA, SOM [e.g., 7-10] and IPS2 [11] an option
is to implement control charts in a form of a service [12-
16], to ensure correctness and a clear definition of SPC
role within a manufacturing system.
provider p
with the objective of changing the conditions under
17]. Use of services
brings external knowledge, expertise and abilities.
Service providers are usually specialized and can
therefore perform requested actions better than the
beneficiary himself.
is Fig. 1. Functional diagram of a service unit [12]
SOA is based on information communication
technologies and mainly uses internet-based
technologies to support the integration of systems that
offer and use services [18].
Huang et al. [19] propose a product service system in
which various information services support production
along a supply chain. Furthermore, in [10], this approach
is extended to material-related services, and called
manufacturing oriented services (MOSs).
Building o 2]
present the functional diagram of a basic service unit,
shown in Fig. 1.
A service unit creates new value by providing
services. Together with service consumers, service units
form a manufacturing oriented service network, where
various services can be consumed by work systems or
other services according to a set of rules and
conventions. A detailed explanation can be found in
[12].
In the next section, an example of a service unit for
SPC is described.
3. Distributed SPC system: eSPC
This Section presents the distributed electronic SPC
(eSPC) concept and ways of its realization. eSPC is a
web service and a corresponding web application,
providing SPC functionality through generation of
control charts and other SPC reports. The role of eSPC
in operations support is illustrated in Fig 2.
The input for eSPC is process data, obtained by a data
acquisition system, and quality related data such as
target values and specification limits, determined by an
expert. The output is control chart and other SPC data in
XML from.
 Gašper Škulj et al. / Procedia CIRP 7 (2013) 401 – 406 403

Permission Organisation Process

eSPC
SERVICE UNIT PermissionID OrganisationID MachineToolID
WORKPIECE/ OrganisationID OrgName ToolID
SPC DATA UserID MaterialID
PROCESS
(XML) WorkOrder
DATA
OrganisationID
MONITORING User
AND CONTROL Measurement Characteristic
UserID
MeasurementID CharacteristicID
CONTROL Password
CharacteristicID MachineID
SENSINGX FEEDBACK SENSINGY Value CharName
Date CenterLine
INPUTS {X} OUTPUTS {Y} Comment StDev
PROCESS
LCL
LEGEND: UCL
PRIMARY KEY UCLr
FOREIGN KEY EwmaLCL
Fig. 2. eSPC role in manufacturing operations support EwmaUCL

A functional overview of eSPC usage is presented in

Fig. 4. eSPC minimal database schema
Fig. 3, which shows how eSPC transforms raw process
data into control charts viewable by the user. The
process consists of three steps: (1) data transfer, (2) data Data processing
processing, and (3) display of SPC data.
In the data processing step process data is
Data transfer transformed into SPC data. The transformation consists
of calculation of statistics for each sample, and
In the data transfer step, production data is sent to application of rules which determine whether the
eSPC through web service methods. Methods are statistics of samples are within limits and thus in control.
available for specification of a new part, specification of In addition to the limit checking, rules (e.g. Western
a new measured characteristic, and insertion of a new Electric rules [20]) are applied to identify out of control
measurement. All data is stored in a database, which samples.
resides on the web server. SPC data is sent back to the user in form of XML
The minimal required database schema is shown in reports. XML structure for control charts is shown in
Fig 4. The schema is based on the following concept. A Fig. 5.
manufacturing organization performs several processes. A control chart report specifies a name generated
For each process, several characteristics can be from product and characteristic data, a centre line along
observed. Each characteristic consists of a set of with control limits, and includes an array of data points,
measurements and SPC related data such as control each associated with a value and a Boolean variable
limits. Each user of the eSPC service has individually specifying whether the data point is out of control. This
granted permissions to access data of an organization. is all the data that is required to draw and display a
control chart.
WORKPIECE/
PROCESS WEB SERVICE DATA IN <ControlChartReport>
DATA METHODS DATABASE <Name></Name>
SPC <CenterLine></CenterLine>
DATA ALGORITHMS SPC <LowerControlLimit></LowerControlLimit>
TRANSFER DATA (XML)
<UpperControlLimit></UpperControlLimit>
DATA DISPLAYED SPC DATA <Data>
PROCESSING <DataPoint>
DISPLAY OF <Value></Value>
DATABASE WEB SPC DATA <OutOfControl></OutOfControl>
(ON eSPC SERVER </DataPoint>
SERVER)
</Data>
WEB SERVICE WEB BROWSER
<ControlChartReport>

Fig. 3. Functional diagram of eSPC

Fig. 5. XML structure of a control chart report
404 Gašper Škulj et al. / Procedia CIRP 7 (2013) 401 – 406

Display of SPC data
Finally, the data is displayed in a web browser by a
web application. The web application consists of several
screens:
a control panel, showing an overview for a
chosen process,
control charts, showing detailed charts for a
characteristic,
histograms, probability diagrams and statistics
screen displaying statistical information and
distribution of the data.
minute changes caused for example by machine wear
that tends to move the process out of control.
Multivariate Hotelling T2 charts take into account
correlation between different measured parameters and
sometimes show that process is out of control even if
control charts for individual parameters do not detect the
change.
Control limits can be calculated from sample points,
or are predefined. Furthermore, historical data can be
viewed from control charts screen by defining a time
window to be observed.

Control panel

The control panel is the main screen of the

application. For a chosen process, all control charts of its
characteristics are shown in one place. The colouring of
the charts is similar to that of traffic lights; green charts
signify that the characteristic is in control, yellow ones
that the characteristic exceeds warning limits, and red
ones that the characteristic is out of control. In this way
the operator is alarmed in a transparent manner simply
by glancing at the control panel.

Control charts

The control charts screen displays detailed control

charts for a chosen characteristic. Samples which are out
of control are clearly marked in red. Several well
established types of control charts are provided i.e. x-R,
EWMA, CUSUM, Hotelling T2. Fig. 7. The Control charts screen

Other screens

Other screens include histograms, probability

diagrams, and general statistics of data. Together, they
provide means for analysis of the observed data;
especially for checking whether the data is normally
distributed. This is important, because the control charts
theory assumes normal probability distribution. This
may seem trivial, but when the data is not normally
distributed, SPC alarms may be triggered for improper
reasons.

Fig. 6. The Control panel screen

Different control charts are used simultaneously for a

various real situations. Shewhart's X-R charts are basic

control charts and are effective for detection of large
shifts in process. EWMA control charts are on the other
hand effective for detection of smaller shifts. CUSUM
control charts are used for trend detection and can reveal
Fig. 8. The General statistics screen
 Gašper Škulj et al. / Procedia CIRP 7 (2013) 401 – 406 405

4. Case study

The primary reason for using control charts is quality

management which is often required by business clients.
This includes both production of quality products as well
as generation of appropriate documentation. The study
presents the use of eSPC in a case of thermoplastic tube
production.
The particular manufacturer is a small on-demand
supplier of extruded plastic tubes. A lot of its customers
are beginning to demand documentation about
manufacturing process for specific manufactured items.
The manufacturer decided to satisfy the demand with the
use of SPC and control charts. To ensure documentation
Fig. 10. Control panel showing a process out of control event
quality, traceability, quality analysis possibility, and gain
a competitive edge, the use of the outsourced eSPC was
selected as best solution for the given problem.

ZONE 5 ZONE 4 ZONE 3 ZONE 2 ZONE 1

ZONE 6

T6 T5 T4 T3 T2 T1

PLC

DATA
CLIENT WWW
TRANSFER
APP.
APP.
WIN. SERVICE WEB BROWSER
CLIENT
Fig. 9. Case study: use of eSPC for temperature control
The process of plastic tube extrusion is observed. For
the finished tube to be acceptable, it has to have specific
material properties that can be reached only by
controlled heating during extrusion. The critical target
parameter, ie. the temperature during extrusion, is
specified by the customer. The manufacturer therefore
has defined target temperature points with tolerances
that must to be reached during manufacturing and has to
provide documentation that the material was treated
correctly during tube extrusion process. It is requested
by the customer that all non conformative tubes are
excluded from delivery.
The diagram in Fig. 9 presents the basic structure of
deployed eSPC service for temperature control in four
heating zones and two zones in extrusion area. The
temperatures are measured by a programmable logic
controller (PLC) and forwarded to a personal computer
(PC) that is connected to the internet.
Fig. 11. Hotelling T2 chart showing the event from Fig. 10.
eSPC
The PC runs a program that is capable to connect
WEB SERVICE
SERVER with the eSPC web service. The gathered data are then
transmitted to eSCP and saved in the online database.
Beside SPC, eSPC is also providing data storage and
therefore continuous accessibility of data. On request,
the data is analyzed and control charts are generated.
Online information is displayed in an intuitive and
understandable manner, as shown in Figs. 10 and 11.
Both figures display control charts for temperature in
tube extrusion process. On some charts it can be clearly
seen that at some point a major jump in temperature
occurred. In this case the change in temperature was
caused by the operator.
Fig. 10 shows that not all charts have detected the
change and even if they did it is not marked as process
out of control event in all cases. For this reason the
control panel is designed to show multiple control charts
side by side. In this kind of situations the use of
multivariate Hotelling T2 chart is also a good idea as
shown in Fig. 11.
Information provided by eSPC acts as documentation
for the customer. This kind of documentation can be
viewed anytime and anywhere, as long as the traceability
of products related to the processes is achieved.
Besides process documentation gathering, eSPC
control charts are also used as a real time control
406 Gašper Škulj et al. / Procedia CIRP 7 (2013) 401 – 406
mechanism. A machine operator receives a warning
when process could be going out of control even if he is
not beside the machine. In this way he is able to react
and improve conformity with the requested process
parameters.
5. Conclusion
In the article, successful development and
implementation of SPC as a Service is presented. The
focus of the SPC service is on control charts with
statistical tools and transparent online graphical
representation of results in real time. eSPC is created as
a lightweight and inexpensive building block that can act
as a replacement for paper SPC and documentation in
small or large distributed manufacturing companies.
The case study illustrates how to use the SPC service
in a small company to increase its capabilities in process
quality management and process control. Present and
past information about manufacturing processes are
readily available to the manufacturer and customer from
anywhere and at anytime.
Analysis of the data, gathered from eSPC, can also be
used for production monitoring and management in its
broadest sense. This kind of information is interesting
for process operators, production managers, quality
personnel, production planning, upper level
management, etc.
The presented use of service oriented principle in
manufacturing could be widely implemented in other
areas of manufacturing or manufacturing support in the
future. The key issue here is that competent solutions
and expertise can be provided to manufacturing
companies, which are usually lacking in-depth
knowledge on advanced methods.
Acknowledgements
This work was partially supported by the Slovenian
Research Agency, Grant No. L2-2001. The authors
would like to thank P.P. Plast d.o.o. company and Mr.
code.
References
[1] Montgomery, D.C., 2009. Introduction to Statistical Quality
Control 6th ed., John Wiley & Sons.
[2] Franceschini, F., Galetto, M., Maisano, D.A., Mastrogiacomo, L.,
2011. ISO/TS 16949: analysis of the diffusion and current trends.
Journal of Engineering Manufacture, 225(5), p.735.
[3] Pekovic, S., Galia, F., 2009. From quality to innovation: Evidence
from two French Employer Surveys. Technovation, 29(12), p.829.
[4] Ab Rahman, M.N., Zain, R.M., Nopiah, Z.M., Ghani, J.A., Deros,
B.M., Mohamad, N., Ismail, A.R., 2009. The Implementation of
SPC in Malaysian Manufacturing Companies. European Journal
of Scientific Research, 26(3), p.453.
[5] Garvin, D.A., 1987. Competing in the Eight Dimensions of
Quality. Harvard Business Review, 87(6), p.101.
[6] Colledani, M., Tolio, T., 2012. Integrated quality, production
logistics and maintenance analysis of multi-stage asynchronous
manufacturing systems with degrading machines. CIRP Annals
61, p.455.
[7] Ueda, K., Takenaka, T., Vancza, J., Monostori, L., 2009. Value
creation and decision-making in sustainable society. CIRP
Annals, 58, p.681.
[8] Na, G.., Songzheng, Z., Xiaodi, Z., 2009. Research on the service-
oriented manufacturing model, in Proceedings of the International
Conference on Industrial Engineering and Engineering
Management IEEM 2009, Hong Kong, p.2431.
[9] Wu, X.Q., Yang, M.S., Gao, X.Q., Ba, L., 2011. Conceptual
Model and Operational Framework of Service-Oriented
Manufacturing. Advanced Materials Research, 179-180, p.1177.
[10] Huang, G., Qu., T., Zhong, R.Y., Li, Z., Yang, H.D., Zhang, Y.F.,
Chen, Q.X., Jiang, P.Y., Chen, X., 2011. Establishing productino
service systems and information collaboration platform for mold
and die products. The International Journal of Advanced
Manufacturing Technology, 52(9), p.1149.
[11] Meier, H., Roy, R., Seliger, G., 2010. Industrial Product-Service
Systems IPS2. CIRP Annals, 59(2), p.607.
[12] Zupa A service network for the
support of manufacturing operations. International Journal of
Computer Integrated Manufacturing, 25(9), p.790.
[13] Cusumano, M., 2010. Cloud computing and SaaS as new
computing platforms. Communications of the ACM, 53, p. 27
[14] Wu, XQ, Yang, MS, Gao, XQ, Ba L., 2011. Conceptual model
and operational framework of service-oriented manufacturing.
Advanced Materials Research, 179-180, p.1177.
[15] Mahdavi, I., Shirazi, B., Cho, N., 2007. A framework of e-based
quality management for distributed manufacturing system.
Contemporary Management Research, 3, p.103.
[16] Sahebjamnia, N., Mahdavi, I., Cho, N., 2010. Designing a new
model of distributed quality control for sub-assemble products
based on the intelligent web information system. Journal of
Intelligent Manufacturing, 21, p.511.
[17] Tomiyama, T., 2001. Service engineering to intensify service
contents in product life cycles, in Proceedings of the Second
International Symposium on Environmentally Conscious Design
and Inverse Manufacturing, Tokyo, Japan, p.613.
[18] Cummins, F.A., 2009. Enabling the Agile Enterprise. Burlington:
Morgan Kaufmann.
[19] Huang, S., Zeng, S., Fan, Y., Huang, G.Q., 2011. Optimal service
selection and composition for service-oriented manufacturing
network. International Joutnal of Computer Integrated
Manufacturing, 24(5), p.416.
[20] Western Electric Company, 1956. Statistical Quality Control
handbook (1 ed.), Indianapolis, Indiana: Western Electric Co..